Skip to content
Other

EVA NSFW-классификатор

четыре уровня вместо бинарного флага

1 min readupdated

Это vision transformer на архитектуре EVA, дообученный для классификации изображений по четырём уровням: neutral, low, medium, high. Модель на ~90 млн параметров обучена на 100 000 синтетически размеченных картинок и может работать как бинарный фильтр или выдавать вероятности по всем классам. Подходит для модерации и фильтрации визуального контента.

Ключевые особенности

  • Четыре класса вместо «NSFW/не NSFW»: neutral, low, medium, high.
  • Обучение на 100 000 синтетически размеченных изображениях.
  • По внутренним бенчмаркам обходит Falconsai/nsfw_image_detection и AdamCodd/vit-base-nsfw-detector.
  • В BF16 на RTX 3090: около 10 мс на изображение при batch 1 и подаче torch tensor.
  • Есть готовый pip-пакет и вариант через transformers pipeline; для продакшена советуют torch tensors.

Пример

from transformers import pipeline
from PIL import Image

classifier = pipeline('image-classification', model='Freepik/nsfw_image_detector', device=0)
image = Image.open('path/to/image.jpg')
print(classifier(image))
ПодробнееГлобальная точность: high — 99.54%, medium — 97.02%, low — 98.31%, neutral — 99.87%.
На RTX 3090 в BF16: batch 1 — около 10 мс и 540 МБ VRAM при подаче torch tensor; batch 16 — около 102 мс и 1144 МБ.
Для продакшена не рекомендуют pipeline: быстрее и удобнее torch tensors; модель обучена в BF16.
Порог low/medium/high стоит подбирать под задачу: на пограничных уровнях возможны ошибки.
Discuss Telegram 0 25

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.2 GB · ready files0.7 GB
  • FP8 / INT8recommendedweights 0.1 GB0.6 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # bakisae/nsfw_image_detector
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download bakisae/nsfw_image_detector --local-dir ~/models/nsfw_image_detector
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to nsfw_image_detector
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of eva02_base_patch14_448.mim_in22k_ft_in22k_in1k (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 2All on Hugging Face
Fine-tunes 7All on Hugging Face

About the model

bakisae/nsfw_image_detector
Author
bakisae
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

250

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments